Reproducible analysis reports with eye-tracking reading time data
  • D. Palleschi
  1. 3  Eye-tracking during reading
  • 1  Course Intro
  • 2  Reproducible Analyses
  • 3  Eye-tracking during reading
  • 4  Working with eye-tracking reading data in R

Table of contents

  • 4 Eye-tracking
    • 4.1 Eye movements
    • 4.2 The eye-tracker
  • 5 Eye-tracking during reading
    • 5.1 Eye-tracking reading measures
    • 5.2 Region of interest (ROI)
    • 5.3 Measures (dependent variables)
    • 5.4 Independent variables
    • 5.5 What do these measures tell us?

3  Eye-tracking during reading

What can we learn from the measures?

Author
Affiliation

Daniela Palleschi

Humboldt-Universität zu Berlin

Published

July 7, 2023

4 Eye-tracking

  • in (psycho)linguistics
    • during reading
    • visual world paradigm
  • in psychology
    • pupillometry
    • visual search
  • but also
    • market research
    • diagnostic tool

4.1 Eye movements

  • saccades: eye movements (e.g., from one word to another)
    • average saccade legnth: 7-9 letters (in alphabetic writing systems)
  • fixations: ‘looking at’ something, e.g., a word (little movement)
    • when information is taken in
    • average duration: 225-250ms (ranging 50-600ms)
  • regressions: saccades to earlier text
    • occurance: 10-15% of saccades in skilled readers

Rayner (2009)

4.2 The eye-tracker

  • eye-tracker
    • camera \(+\) infrared illuminator
  • screen
  • chin/head rest
  • in our lab: desk-mounted

Image source: SR Research (all rights reserved)

5 Eye-tracking during reading

5.1 Eye-tracking reading measures

  • inform theories of language processing via linking hypotheses
    • linking visual attention to processing
  • typically, we compare reading times as a function of some manipulation
    • e.g., Sally went/goed to the store.
  • longer reading times are taken to reflect processing costs, associated with e.g., sentence complexity or anomalies

5.2 Region of interest (ROI)

  • can be anything on-screen
    • sentence-level
    • word/region-level
    • a certain part of the screen

Video

5.3 Measures (dependent variables)

  • what we measure = dependent variables (usually…)
    • their value depends on some predictor (e.g., word frequency)
  • measures of duration (time spent on a region)
    • first fixation
    • first-pass reading time
    • regression path duration
    • total reading time
  • data type: continuous
  • measures of revisits
    • number of fixations
    • number of regressions in/out
    • regression in/out (yes or no)
    • probability of regressions in/out (0:1)
  • data type: binary (0,1) or count

5.4 Independent variables

  • what can influence reading measures? (Clifton & Staub, 2011; Juhasz & Pollatsek, 2011; Rayner & Liversedge, 2011; Warren, 2011)
    • some examples:
  • Word properties
    • word frequency
    • word length
  • Sentence-level influences
    • context (i.e., prediction)
    • semantic or grammatical manipulations
  • Inter- and intra-individual
    • domain-specific expertise
    • reading skill level

5.5 What do these measures tell us?

  • eye-tracking during reading can tell use when and where processing costs are incurred
  • early measures involve “first contact with a word” or region: first-fixation, first-pass reading time (Vasishth et al., 2013, p. 126)
  • late measures involve regressions to a region: e.g., total reading time
    • may also include ‘spillover’ effects from early processing
  • eye-tracking during reading measures can therefore tell us about stages of processing

References

Clifton, C., & Staub, A. (2011). Syntactic influences on eye movements during reading. In Oxford Handbook of Eye Movements. Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199539789.013.0049
Juhasz, B. J., & Pollatsek, A. (2011). Lexical influences on eye movements in reading. In Oxford Handbook of Eye Movements.
Rayner, K. (2009). Eye movements and attention in reading, scene perception, and visual search (Vol. 62, Issue 8). https://doi.org/10.1080/17470210902816461
Rayner, K., & Liversedge, S. P. (2011). Linguistic and cognitive influences on eye-movements during reading. In Oxford Handbook of Eye Movements.
Vasishth, S., von der Malsburg, T., & Engelmann, F. (2013). What eye movements can tell us about sentence comprehension. Wiley Interdisciplinary Reviews: Cognitive Science, 4(2), 125–134. https://doi.org/10.1002/wcs.1209
Warren, T. (2011). The influence of implausibility and anomaly on eye movements during reading. In Oxford Handbook of Eye Movements.
2  Reproducible Analyses
4  Working with eye-tracking reading data in R
Source Code
---
title: "Eye-tracking during reading"
subtitle: "What can we learn from the measures?"
author: "Daniela Palleschi"
date: "`r Sys.Date()`"
institute: Humboldt-Universität zu Berlin
lang: en
bibliography: references/references.json
csl: references/apa.csl
---

```{r, eval = T, echo = F}
knitr::opts_chunk$set(eval = T, # evaluate = T for REPRODUCIBLE analyses
                      echo = F, # 'print code chunk?'
                      message = F, # print messages?
                      error = T, # render even if errors encountered?
                      warning = F) # print warnings?
```

```{r, eval = T, cache = F, echo = F}
# Create references.json file based on the citations in this script
# make sure you have 'bibliography: references.json' in the YAML
rbbt::bbt_update_bib("_et_reading.qmd")
```

# Eye-tracking

- in (psycho)linguistics
  + during reading
  + visual world paradigm
- in psychology
  + pupillometry
  + visual search
- but also
  + market research
  + diagnostic tool

## Eye movements

- **saccades**: eye *movements* (e.g., from one word to another)
  + average saccade legnth: 7-9 letters (in alphabetic writing systems)
- **fixations**: 'looking at' something, e.g., a word (little movement)
  + when information is taken in
  + average duration: 225-250ms (ranging 50-600ms)
- **regressions**: saccades to earlier text
  + occurance: 10-15% of saccades in skilled readers

::: {.footer}
@rayner_eye_2009-1
:::

## The eye-tracker

:::: {.columns}

::: {.column width="50%"}
- eye-tracker
  + camera $+$ infrared illuminator
- screen
- chin/head rest
- in our lab: desk-mounted

:::

::: {.column width="50%"}
```{r eval = T, fig.env = "figure",out.width="100%", fig.align = "center", set.cap.width=T, fig.cap="Image source: [SR Research](https://www.sr-research.com/about-eye-tracking/) (all rights reserved)"}
knitr::include_graphics(here::here("media/SR_Research.png"))
```

:::

::::

# Eye-tracking during reading

## Eye-tracking reading measures

- inform theories of language processing via linking hypotheses
  + linking visual attention to processing
- typically, we compare reading times as a function of some manipulation
  + e.g., Sally *went*/*goed* to the store.
- longer reading times are taken to reflect processing costs, associated with e.g., sentence complexity or anomalies

## Region of interest (ROI)

- can be anything on-screen
  + sentence-level
  + word/region-level
  + a certain part of the screen
  
::: {.content-visible when-format="revealjs" when-format="html"}
:::: {.columns}

::: {.column width="60%"}
![](media/crit_trial_px03.mp4){fig-align="center" width="100%"}
:::

::::
:::

## Measures (dependent variables)

- what we measure = *dependent* variables (usually...)
  + their value *depends* on some *predictor* (e.g., word frequency)

:::: {.columns}

::: {.column width="50%"}

- measures of duration (time spent on a region)
  + first fixation
  + first-pass reading time
  + regression path duration
  + total reading time
- data type: *continuous*
  
:::
 
::: {.column width="50%"}

- measures of revisits
  + number of fixations
  + number of regressions in/out
  + regression in/out (yes or no)
  + probability of regressions in/out (0:1)
- data type: *binary* (0,1) or *count*
  
:::

::::

## Independent variables

- what can influence reading measures? [@juhasz_lexical_2011; @rayner_linguistic_2011; @warren_influence_2011; @clifton_syntactic_2011]
  + some examples:

:::: {.columns}

::: {.column width="26%"}
- Word properties
  - word frequency
  - word length

:::

::: {.column width="37%"}
- Sentence-level influences
  - context (i.e., prediction)
  - semantic or grammatical manipulations

:::

::: {.column width="37%"}
- Inter- and intra-individual
  - domain-specific expertise
  - reading skill level

:::

::::


## What do these measures tell us?

- eye-tracking during reading can tell use *when* and *where* processing costs are incurred
- *early measures* involve "first contact with a word" or region: first-fixation, first-pass reading time [@vasishth_what_2013, p.126]
- *late measures* involve regressions to a region: e.g., total reading time
  + may also include 'spillover' effects from early processing
- eye-tracking during reading measures can therefore tell us about stages of processing

# References {.unlisted .unnumbered visibility="uncounted"}

::: {#refs custom-style="Bibliography"}
:::